Providing alphanumeric input to interactive voice response (IVR) systems tends to be difficult. For example, entering a vehicle's VIN number or even the name of a person using T9 text input in the middle of a telephone call is a challenging experience. Therefore, alphanumeric input is typically provided to an IVR system using speech that is recognized by Adaptive Speech Recognition. However, speech input is not secure for use in public places where there is a chance of being overheard, and speech recognition does also not work well in a noisy environment.
Financial IVR applications require users to enter information such as credit card and bank account numbers, a mother's maiden name, passwords, etc. Some of this secure information might be provided using speech input that renders the IVR application useless when the caller is in a public place (e.g., on a crowded train) where there is a chance of being over-heard. For example, if a person speaks out his credit card number over a phone in a train, then there is a chance that a fellow passenger with a good memory might memorize that number. DTMF input can be used to provide numerical data such as a credit card number securely, but having to enter a long string of digits and correcting for input errors is not a user friendly experience. The problem is especially pronounced for users with a physical handicap.